The Truthful Art — Ramblings re: Chapter 3 & 4
The example provided in Chapter 3 — with the headline “study finds more than a quarter of journalism grads wish they had chosen another career” — made me think about my own occasional tendency to digest headlines rather than seek to understand the whole story. We are, now more than ever before, overwhelmed by the amount of information at our fingertips. Which has seemingly led to organizations seeking to minimize the “story” into a single, easily digestible takeaway. They are doing what Cairo cautions against when referencing John Maeda’s creed about simplicity — subtracting the obvious, without always adding the meaningful.
But the fault doesn’t always lie only with the organizations that put forward models. I think back to the polling and subsequent predictions of the 2016 presidential election. Nate Silver’s fivethirtyeight.com routinely described the polling situation in terms of probabilities. You could view the model on their sight, see their methodology, and see how they came to a reasonable conclusion that Hilary Clinton was a clear favorite to win the election. When that didn’t happen, a lot of people, including some news media, decried “the polls were wrong! We shouldn’t trust the polls! Nate Silver was wrong!” They didn’t care to consider sampling error (which fivethirtyeight often discusses) or the realities of a probability-based model. Unfortunately, a lot of people didn’t care about the rigorous attempts to make the model “more truthful” as Cairo describes in his book. Instead, it was used as an opportunity to say that the data was a lie (it wasn’t) and that polls should never be trusted (unless they say what I want them to say). It serves as a stark reminder that this field is about far more than making beautiful illustrations and engaging visualizations. We also need a fundamentally better understanding of data, and sadly, a better understanding of what it means to be “true.”
A brief aside to close — Chapter 4 of The Truthful Art is one of the better “layman’s terms” explanation of the scientific method, variables, and error. Having completed most of a quant-focused graduate program, none of this chapter was particularly new to me. But it was thorough without becoming overwhelmingly technical. I have been asked to teach an undergraduate statistics course in the spring semester. Given that I anticipate the vast majority of the students enrolled will want to be anywhere else on earth other than in that class, I think I may lean on The Truthful Art for some explanations of the key topics. Particularly the different types of variables — nominal, ordinal, interval, and ratio — which I know often trip people up.
